Classification of engine lubricant oil degradation using electronic nose / Suhaimi Mohd Daud

Faculty of Electrical and Electronics Engineering

Thesis (Master of Science) -- Universiti Malaysia Pahang – 2018

Includes bibliographical references

Lubricating oil is one of the products of petroleum refinery and it is one of the largest areas of research in the automotive sector. It is very useful in maintaining optimum engine performance in order to protect the engine from the effects of heat, pressure, corrosion, oxidation, contaminants, and cooling of an engine caused by two moving parts that touch each other. The aim this study is to classify the lubricant oil sample based on its odor using electronic nose as the main instrument and classify the lubricant oil odor-profile using case-based Reasoning and K-Nearest Neighbor (K-NN) classifier. There are several methods used to study and analyze the lubricant oil such as ICP-MS, Spectrophotometer and e-nose. E-nose is one of the approaches in analyzing lubricating oils by using odor changes as parameters. Through the smell, lubricating oils can be analyzed based on the changes in odor caused by changes in the content and effects of various factors for each lubricating oil sample. In this study, e-nose has been used to measure and collect odor raw data for the lubricant oil sample comprising 11 oil samples which are 1 sample from virgin oil and 10 samples from oil that have been used at 1000KM up to 10000KM of the engine. The raw data collected will be through a signal-processing phase with the aim of extracting the odor-profile for each lubricating oil sample. This odor-profile is a very useful pattern for the process of classification using Case-based Reasoning (CBR) and K-Nearest Neighbor (K-NN). Furthermore, this odor-profile is the combination of several unique features that produce a very significant odor pattern for each sample. Hence, the meaningful odor patterns that represent the odor-profile of each oil sample is the factor that makes the process of classification become very good and significant. Next, these meaningful odor-profiles of each sample were classified using the CBR and K-NN classification technique in order to differentiate between a sample and another based on the odor that produced by the each sample. Both of these techniques are very popular and widely used in the field of classification. The results of this study show that the classification odor profile for each sample lubricant has successfully achieved 100% classification using CBR classifier.